I was wondering if there is any documentation (or possibility) to build the older (historic) versions of reduce. There are quite a number of packages published (e.g., in Computer Physics Communications) that do not work with the current Reduce versions (e.g., RicciR, Supercalc).
With supercalc, for example, where Reduce gave an error message when trying to compile it, that I fixed easily just out of curiosity, but it still does not give correct results.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The older versions are there just for interest and as a "historical
record". Back at the time they existed they tended to be built mostly be
"specialists" either on the PSL or CSL side.. and certainly that will have
pre-dated the autoconf stuff that now attempts (in a fairly ugly way!) to
make builds "just happen" on all platforms.
There is no neat consolidated list of what has changed either!
I would expect that most older code would have at worst rather simple
issues when built using the Reduce variant that its developers used, in
that neither CSL nor PSL has changed that radically... even though all
sorts of small details will have moved forward. In cases where the authors
could be contacted and were willing to license their code suitably any of
these that were of interest could be saved in the reduce "contrib"
directory while still in an uncartain state, and then if some helpful
volunteer was willing to sort out any oddities and ensure that there was a
suitable bit of documentation and a test file they could migrate into the
main tree and be built at part of the regular system.
Building the ancient versions would almost certainly be possible on at
least ons platform, but I think it would be much more productive to get
packages made freely available to all and then eventually set up so they
do not need to be built individually by hand but they just form part of
the full system.
Arthur
On Thu, 7 Mar 2019, arpi wrote:
I was wondering if there is any documentation (or possibility) to build
the older (historic) versions of reduce. There are quite a number of
packages published (e.g., in Computer Physics Communications) that do not
work with the current Reduce versions (e.g., RicciR, Supercalc).
With supercalc, for example, where Reduce gave an error message when
trying to compile it, that I fixed easily just out of curiosity, but it
still does not give correct results.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I was wondering if there is any documentation (or possibility) to build the older (historic) versions of reduce. There are quite a number of packages published (e.g., in Computer Physics Communications) that do not work with the current Reduce versions (e.g., RicciR, Supercalc).
With supercalc, for example, where Reduce gave an error message when trying to compile it, that I fixed easily just out of curiosity, but it still does not give correct results.
The older versions are there just for interest and as a "historical
record". Back at the time they existed they tended to be built mostly be
"specialists" either on the PSL or CSL side.. and certainly that will have
pre-dated the autoconf stuff that now attempts (in a fairly ugly way!) to
make builds "just happen" on all platforms.
There is no neat consolidated list of what has changed either!
I would expect that most older code would have at worst rather simple
issues when built using the Reduce variant that its developers used, in
that neither CSL nor PSL has changed that radically... even though all
sorts of small details will have moved forward. In cases where the authors
could be contacted and were willing to license their code suitably any of
these that were of interest could be saved in the reduce "contrib"
directory while still in an uncartain state, and then if some helpful
volunteer was willing to sort out any oddities and ensure that there was a
suitable bit of documentation and a test file they could migrate into the
main tree and be built at part of the regular system.
Building the ancient versions would almost certainly be possible on at
least ons platform, but I think it would be much more productive to get
packages made freely available to all and then eventually set up so they
do not need to be built individually by hand but they just form part of
the full system.
Arthur
On Thu, 7 Mar 2019, arpi wrote: